A modular agent-based framework for wildfire evacuation: Integrating fire spread, multimodal transport, and dynamic routing

• This study presents a modular agent-based framework coupling wildfire spread with multimodal evacuation. • Fire-driven road-segment closures dynamically update network availability during the evacuation. • A custom waypoint-based routing algorithm enables adaptive rerouting for outbound and inbound evacuation. • Pedestrian movement, private vehicle, emergency extraction, and vehicle–pedestrian interactions are explicitly represented. • The model was applied to a WUI and validated against an evacuation drill to assess realism and operational consistency. Wildfires increasingly threaten communities, especially in the Wildland–Urban Interface, where evacuation often becomes the only viable life-safety measure. Planning effective evacuations is challenging due to interactions among community behavior, road network capacity, and rapidly changing fire conditions. Despite advances in wildfire evacuation modeling, significant gaps remain: limited dynamic rerouting when roads are compromised, insufficient representation of multimodal evacuation, weak integration of high-resolution fire spread data, and the lack of inbound traffic and rescue operations. We introduce a modular, agent-based evacuation framework that combines fire spread, pedestrian movement, and vehicle traffic in a single model. It supports dynamic routing responsive to advancing fires, enables both outbound self-evacuation and inbound rescue operations, and incorporates multimodal transportation modes. A high-resolution wildfire model is coupled with diverse human behaviors and route availability. The model was tested in a case study in Portugal and validated against an evacuation drill, showing strong agreement in evacuation times and demonstrating its capacity to replicate real-world processes. Owing to its modular design and use of publicly available datasets, the framework can be readily adapted to diverse settings. It improves evacuation planning by allowing scenario testing, “what-if“ analyses, and data-driven decision-making to enhance community safety and resilience.